The Effects of Index Sampling on Fixed Income Exchange-Traded Fund Tracking Error

Abstract

This research study examines the impact that index sampling may have on the tracking error for passively managed fixed income exchange-traded funds (ETFs). Tracking error measures the degree to which an ETF's return differs from the return of its underlying index.

The majority of fixed income ETFs issued in the U.S. are managed by representative sampling of their underlying indices. This is because the typical fixed income index is (a) made up of a substantial number of different issues that are constantly changing, and (b) not all issues are readily available for purchase by an ETF manager. As a result, ETF managers must sample the issues in each index to attempt to minimize tracking error and provide returns to investors representative of their underlying index returns.

This study examines whether the degree of sampling is a factor above and beyond expense ratio as a component underlying tracking error. Along with examining the overall fund effects in the sample, this study examines whether characteristics of the underlying benchmark such as credit quality demonstrate different degrees of tracking error as a result of the sampling strategy. Finally, larger firms that have a significant number of fixed income ETF products are examined for the degree of tracking error. The underlying theory behind this examination is that these larger firms have more sophisticated trading operations that are better equipped to buy and sell more representative securities to help manage tracking error.

The model results suggest that consistent with prior research, the expense ratio is a significant variable that adds to tracking error while ETFs composed of mostly investment-grade credit securities exhibit significance that diminishes tracking error. Variables measuring the percentage of securities owned relative to their benchmarks as well as firm size (as signified by the four largest issuers) were found not to be significant. The results suggest that the more securities in an underlying benchmark, the greater the difficulty a manager will have controlling tracking error. The significance of the number of securities in a benchmark suggest that this information should have a more prominent role in required disclosures.